Uncle Sam Buys an Airplane

How Lockheed Martin beat Boeing for the biggest military contract in history—and how that one contract could change the way the military builds and pays for its weapons

A few minutes before two o'clock on a Friday afternoon last fall, Dain Hancock stood up and left a packed auditorium in Fort Worth, Texas. He tried to be unobtrusive, but every eye in the room followed him. The 500 people gathered there, plus thousands more watching closed-circuit broadcasts in other cities, were employees or associates of the Lockheed Martin corporation, all waiting to hear whether their company had won the largest military contract ever awarded. The competition for this contract, to build a new airplane called the Joint Strike Fighter (JSF), had begun almost a decade before. The orders a contract would produce could provide jobs for decades, beyond the likely retirement age of everyone in the room. Pentagon representatives had agreed to telephone the heads of the competing companies just before the official announcement was made. Hancock, the president of Lockheed Martin's aeronautics division, was on his way to get the news from his boss.

Five minutes later he was back, stone-faced. He took a seat in the front row of the auditorium near Tom Burbage, a former Navy test pilot who directed the Lockheed Martin team that was competing for the contract. Burbage leaned over and asked, "What happened?" "Vance called," Hancock whispered, referring to Vance Coffman, Lockheed Martin's CEO. "We won. But don't react."

Burbage, a lanky, athletic-looking man in his fifties with a long jaw, stared ahead, expressionless. The live broadcast from the Pentagon began, and for ten minutes officials seemed to make a game out of delaying the information everyone wanted to hear: whether the JSF contract would go to the team at Lockheed Martin or to a competing team at Boeing. Finally James Roche, the Secretary of the Air Force, took the podium to announce the winner. "Both proposals were very good," he said. "Both demo programs were very good. But on the basis of strengths, weaknesses, and degrees of risk of the program, it is our conclusion ... that the Lockheed Martin team is the winner of the Joint Strike Fighter program on a best-value basis."

Before the word "Lockheed" was fully out of Roche's mouth, the audience in Fort Worth erupted and drowned out the rest of his remarks. Senator Kay Bailey Hutchison, a Republican from Texas who had come to show her support, leaped into the air and screamed. So did Kay Granger, a former mayor of Fort Worth and also a Republican, who now represented the district in Congress. Burbage was one of the few who stayed in their seats—a display of dutifulness for which he was ribbed at the company parties that went on through the weekend.

The joy in Fort Worth was matched by despair in St. Louis, in Seattle, in Washington, D.C., and in other places where Boeing employees and associates had gathered to watch the broadcast. The last time a competitor had been eliminated from the JSF contest, it had gone on to lose its corporate independence. That was in 1996, when the field of candidates shrank to two in a process known at the Pentagon as a "downselect." The renowned fighter-plane maker McDonnell Douglas, nicknamed "McAir," failed to make the cut. Within a year McAir had merged with Boeing, under terms that replaced the McDonnell Douglas name with Boeing's on the factories in St. Louis where McAir's F-4 Phantom, F-15 Eagle, and F-18 Hornet were made.

Boeing was big and diversified enough not to fear going out of business without the JSF, which will be known as the F-35. But Lockheed Martin depended more heavily on defense contracting, and had more at stake. For twenty-five years its mile-long factory in Fort Worth had kept busy producing the F-16 Falcon fighter plane for the U.S. Air Force and more than a dozen foreign militaries. The F-16 (originally made by a division of General Dynamics that is now part of the Lockheed Martin conglomerate) was nearing the end of its long and highly successful run. If Lockheed Martin could not follow it with the JSF, its officials warned, the company would be consigned to the parts business, surviving on subcontracts from Boeing and other strong manufacturers. Boeing's officials had viewed this as poormouthing—part of an attempt to build sympathy for the company and tilt the JSF decision in Lockheed Martin's favor. But the impact on both companies was clear soon after the decision was announced.

In Fort Worth, Lockheed Martin began clearing room on the factory floor for the new JSF production line. It began taking bids from suppliers and subcontractors, it made presentations to foreign governments that might eventually buy the airplane, and it hired new employees at a rate of 250 a month. In St. Louis, Boeing immediately canceled plans to build a new, modern factory for JSF production. It removed the posters and models for its version of the JSF from its buildings, and began looking for other roles for the hundreds of people on its JSF team. In Washington its lobbyists redoubled their efforts to shore up Boeing's defense work. These included supporting a controversial and successful congressional campaign to get the Air Force to lease Boeing civilian airliners for conversion to military tankers, while Missouri politicians launched a controversial and unsuccessful effort to force Lockheed Martin to share the work with Boeing and build some of the new JSFs in St. Louis. Early this year the victorious Lockheed Martin sent a team of recruiters to set up a job fair in a motel near Boeing's plant in St. Louis. The recruiters placed ads in the local papers, encouraging Boeing employees to switch teams and be part of "the future of tactical aviation!"

Boeing's response was to emphasize what it considered to be the real future of tactical aviation—unmanned combat and reconnaissance aircraft that were more-advanced versions of the Predator drones used in Afghanistan. In formal and informal comments to military officers, civilian analysts, politicians, consultants, reporters, and other members of the defense community, Boeing representatives were careful never to speak dismissively of the JSF. But the next great advance, the one that mattered, would be in unmanned combat vehicles. These could be built more cheaply than normal airplanes, because they could omit many systems designed for pilot protection, and they could save American lives. In this field Boeing held the edge.

Beyond its impact on the two companies, the JSF project is enormously important for the U.S. military—and for U.S. taxpayers, who in the next few years will be asked to support this and other plans that will make for the largest increase in defense spending since the height of the Cold War. The JSF matters because of both its scale and its conceptual ambition. The planners at the Pentagon and at Lockheed Martin imagine that as many as 6,000 of these airplanes may be bought, at a total cost of as much as $200 billion, over the next twenty-five years. If all goes according to plan, about 3,000 of the JSFs will go to the original "investors" in the program—the U.S. Air Force, Navy, and Marine Corps, plus the Royal Air Force and Royal Navy in Britain. All have shared the cost of developing the plane. The other 3,000 are supposed to go to customers in the rest of the world. Of the 4,000-odd F-16s produced since the late 1970s, nearly half have gone to foreign customers, and the idea is that the JSF will be even more attractive as an export airplane. "If it succeeds, it will simply dominate the world market," Richard Aboulafia, a well-known aviation analyst, says. "It could do to the European fighter industry what the F-16 almost did: kill it."

The ambitious idea behind the JSF is to address several chronic problems of U.S. military acquisition policy simultaneously. If it succeeds, it will put military procurement on a more affordable, more effective track. If it fails, it will underscore how deep those problems are.

The most obvious novelty of the JSF is the concept that one airplane can fill the needs of three military branches. Officially, the Air Force, the Navy, and the Marines make up only two services, because the Marine Corps is part of the Navy. But the Marines differ greatly from the rest of the Navy in operational requirements, and they have zealously fought to preserve the independence of Marine Corps aviation, with airplanes designed to support them in ground combat.

Each service, in fact, differs in its conception of the ideal fighter plane. The Air Force traditionally yearns for whatever is fastest and most elaborately equipped, in order to gain air superiority against enemy fighters. The Navy needs fighters that are built like trucks, to survive the shock of landing on carrier decks and the constant exposure to salt air, far from maintenance depots. The Marine Corps wants airplanes designed for expeditionary flexibility—planes that can follow the Marines into battle and operate from small, remote landing strips. Because of these differing needs, to say nothing of normal service parochialism, each of the three branches has preferred to design and buy its own equipment. The modern history of joint aircraft for the U.S. military is dominated by one outright disaster—the notorious TFX project of the early 1960s, which led to an expensive fighter that neither the Navy nor the Air Force wanted to use—and one unsatisfying success. This was the F-4 Phantom, used extensively in Vietnam. The F-4 was conceived and built as a Navy airplane, which civilian officials then obliged the Air Force to buy as well. Because of its carrier heritage, it was far heavier and therefore less maneuverable than an Air Force-only fighter might have been.

The JSF project undertook to change this. Its gamble was that a family of closely related airplanes could be different enough to meet each service's needs and yet similar enough to realize major economies of scale. Automobile and computer companies had been pulling off this feat for years: an SUV and a pickup truck appeal to different customers but use the same engine, drive train, chassis, and so on. Applying this logic to military equipment, with similar engines, instruments, and software, would be a first.

The JSF was also novel in its approach to global industrial policy. The F-16, still the reigning export champ, was designed for U.S. purposes and then sold around the world. For both military and commercial reasons, the JSF was conceived as a "world airplane." The military advantage of this approach is "interoperability": in some future joint effort like the Gulf War or the battle in Kosovo, airplanes from allied nations would have the same parts, the same software, the same repair kits. (The Pentagon is now working out rules for the "tailorability" of the JSF, which will presumably include limited software or less advanced radar systems.) The commercial advantage was like that of other enterprises in the global era: being able to attract customers and build supply networks around the world. Starting with Britain, other countries were invited to share the development costs of the airplane and in return to have their companies become eligible for competition as suppliers. Rolls-Royce now makes a crucial propulsion-system part in the winning Lockheed Martin design, and BAE, the former British Aerospace, will make part of the airframe. Holland, Denmark, Italy, Turkey, and Norway, among others, are seen as likely participants.

The real ambition of the project involves cost. The inability to control the cost of modern weapons is arguably America's greatest long-term military vulnerability. Or if not quite the same as openness to terrorism or rogue-state nuclear attack, it is a huge handicap in our ability to respond.

The jokey way of stating this problem is as one of "Augustine's Laws," which were formulated by Norman Augustine, a former chairman of Martin Marietta. His Law XVI was "In the year 2054, the entire defense budget will purchase just one aircraft. This aircraft will have to be shared by the Air Force and Navy 3 1/2 days each per week, except for leap year, when it will be made available to the Marine Corps for the extra day." The serious truth behind this law is that since the end of World War II the cost per weapon for airplanes, tanks, and ships has risen much faster than the defense budget has, so every single category of weapon has declined in numbers. The United States built some 14,000 P-51 and 15,000 P-47 fighters during World War II. It built more than 6,000 of the main Korean War fighter, the F-86, plus more than 7,000 F-84s. It built 5,200 Vietnam-era F-4 fighters. Of the F-16s, designed to be cheap and numerous, the Air Force acquired just over 2,200. The current high-end fighter is the F-22—and the Air Force now plans to buy about 300. The capabilities of the new equipment do offset the diminished numbers, to some extent. But even in the military there is unease about how far the numbers have fallen—or, to use the military term, how much the force structure has shrunk.

The truly revolutionary idea behind the JSF was to break that pattern, by building a modern airplane that was cheap enough to offset the steady decline in force structure. The project would do this by applying innovations from civilian industry, especially modern "lean manufacturing" techniques, in a way never before seen in the defense world. Even more unusual was the notion that its success would be judged by whether or not it stuck to its projected costs. The target costs, as stated in 1994 dollars, ranged from $28 million per plane for the Air Force version of the JSF to $38 million for the Marine Corps version. Although comparing different planes from different eras is difficult, the target costs would put the JSF in roughly the same cost bracket as the F-16. The F-22, by comparison, will cost $100 million apiece and perhaps much more, depending principally on how many units of the plane Congress ultimately votes to build.

The main reason to think this gamble will pay off is the JSF's record up through the moment when Boeing and Lockheed Martin got the news: the cost of the rival models was surprisingly close to target levels. The main reason to expect failure is the record of most weapons projects in the past half century. If the cost of this airplane rises even half as much as the historical norm, the JSF will not come close to its advertised purpose of replacing today's aging fleet. We are midway through a drama that will help determine what practical results we get for our military investment.

This new airplane has gotten as far as it has because of the cunning, ingenuity, and perseverance of people like Darleen Druyun, George Muellner, and Jacques Gansler—people largely unknown to the public who have fought the battles over the JSF inside corporate, governmental, and military bureaucracies. They have won some crucial victories. The hardest work is still ahead.

Avoiding the Defense Death Spiral

Without intending to be, Dick Cheney was the father of the Joint Strike Fighter. In January of 1991, as the Secretary of Defense, Cheney was wrapped up in preparations for war in the Persian Gulf. But on January 7, ten days before the fighting began, he decided to cancel work on a new Navy airplane called the A-12. The A-12 had become a nightmare case of modern military procurement, with the contractor and the military exchanging accusations of deceit and malfeasance, and with price projections rising above $100 million per plane. Cheney decided to write off hundreds of millions of dollars in start-up costs so as to save billions in future commitments.

Suddenly the Navy had a problem, which in different ways the Air Force and the Marine Corps had too. None of them had financially realistic plans for getting the new airplanes they wanted. The Air Force's situation was especially acute. Starting in the 1970s it had put well over 2,000 new fighter and attack planes into its squadrons, in a "high-low mix." The "high" planes were the costly, complicated F-15s, designed for air superiority over the Soviet Union and everyone else. The "low" planes were the less expensive but in many ways more effective F-16s and A-10s. Because these airplanes had been new all at once, they got old all at once. At the time of Cheney's decision, most of the Air Force's inventory was made up of planes that would reach the end of their projected service near the turn of the century. Because it takes a decade or more to design and produce new airplanes, and because aging fighter planes are steadily more expensive to keep patched up and flying, the Air Force needed to do something. (As for the lumbering B-52 bombers, which were new when George W. Bush was in grade school and yet are still flying over Afghanistan, the operating stress is lower on them than it is on fighter planes. And because the bomber fleet is smaller and unit costs for replacements are much higher, it makes economic sense to keep the bombers flying.)

The one plane the Air Force had in the works clearly wouldn't solve the larger problem. In the 1980s it had begun work on the F-22, the ultimate go-anywhere, fight-anything Cold War weapon. To ensure that the plane could penetrate Soviet air defenses and beat any MiG then on the drawing boards, it was designed with barely an eye to cost. When the Soviet Union and the threat of its air force went away, the F-22 remained—but it entered what Chuck Spinney, an influential budget analyst at the Pentagon, has called the "Defense Death Spiral." In the death spiral Congress and the Pentagon buy into a program because the promised performance is so high that the projected costs seem acceptable. Then problems emerge, the schedule slips, and the promised costs start to seem "unrealistic." (The corollary in civilian life is what happens after you contract to renovate your kitchen.) Congress notices the cost and starts worrying. It slows down the production rate and reduces the total order, so the per-unit price really soars. Then Congress or an administration gets serious about cutting the program.

This is what happened to the B-2 bomber. The Air Force originally planned to buy more than 100 and ended up with twenty-one, at a cost of more than $1 billion apiece. As for the F-22, the first Bush Administration signed up for 750, at $50 million apiece. By the end of the Administration the "buy" had been cut to 648. During the Clinton Administration the buy was reduced, in stages, to 339—on the reasonable grounds that the plane was getting too expensive and that the Soviet threat had faded. The current Bush Administration has lowered the acquisition target yet again, to around 300.

The Air Force leadership has remained loyal to the F-22, consistent with its general enthusiasm for the highest-end, most sophisticated fighter conceivable. "This airplane will make the skies safe for everybody else," was the way a retired four-star Air Force general recently put it to me. But replacing the small F-16s and A-10s with the new airplanes would mean, in the words of Jacques Gansler, who directed defense acquisition policy during Bill Clinton's second term, "cutting the force structure in half." So in 1991 the Air Force was shopping for a cheaper airplane.

The Marine Corps had its own sort of airplane emergency. Marine aviation has had two enduring goals: to retain an independent fleet of aircraft, so that in a ground battle the Corps need not rely on Navy planes operating from aircraft carriers; and to have airplanes suited to the "expeditionary," first-to-arrive tenets of Marine doctrine. The Corps' answer had been the Harrier, a British-designed "jump jet" that can take off and land vertically, like a helicopter. That the Harrier works at all is remarkable. Of the several dozen other vertical-landing airplanes the world's militaries have tried to produce, only the Harrier and an old Soviet model called the Yak-36 ever went into production. But the Harrier was always so close to not working, its jet engine running at redline (the maximum safe power setting) to produce enough thrust to lift it up and down, that its safety record was by far the worst of any airplane in U.S. military use. In any case, the Harriers were older and much creakier than the Air Force's F-16s. The Corps wanted something to take their place, and its representatives told any politician who would listen about the need for a new STOVL ("short takeoff vertical landing") airplane.

Then there was the Navy. Cheney canceled the A-12, which would have given it a new airplane that could fly greater distances from a carrier deck and be protected by stealth (anti-radar) technology, just like the latest Air Force airplanes. The Navy responded by ordering a larger and more expensive version of its F-18 Hornet—the F-18E/F Super Hornet. But this put it in the same bind as the Air Force, because the Super Hornet, like the F-22, was too expensive to buy in large numbers.

This situation became known in the military as the tactical-aviation train wreck. The Air Force was trying to buy more F-22s, the Navy was trying to buy its expensive new airplanes, the Marines wanted something different—and no one imagined that there was enough money to satisfy all their demands. Meanwhile, the existing fleet kept getting closer to the end of its projected service life.

In principle the Air Force could have solved its problem simply by placing new orders for the trusty old F-16. Apart from the instinctive aversion to going back to old equipment, the main argument against this strategy was that the F-16 had been designed before the discovery of stealth technology, and would therefore always be more visible on radar than newer planes.

Cheney passed the cost problem to Les Aspin, the Clinton Administration's first Secretary of Defense. Aspin served through Clinton's first year, resigned after the "Black Hawk Down" debacle in Somalia, and was replaced by his undersecretary, William Perry. (A year and a half later Aspin died, at fifty-six, of a stroke.) Solving the services' aviation problems with the kind of airplanes then available was simply impossible. As part of their "bottom-up review" to determine what kind of military the United States needed and could afford now that the Cold War was over, Aspin and Perry created a project called JAST, for "joint advanced strike technology." JAST was a research effort to determine, among other things, whether technical advances like those from civilian industry could be used to avert the train wreck. The findings amounted to a yes, and in 1994 George Muellner learned that he would be the one to make this happen.

Where No General Had Dared Go Before: The Bottom Line

In 1994 George Muellner was a one-star Air Force general, fifty-one years old. He had been a decorated fighter pilot in Vietnam and a commander during the Gulf War. He was on his way to what he considered a plum assignment in the Pentagon when he got a call from Merrill McPeak, then the Air Force Chief of Staff. McPeak asked Muellner how he would like to be in charge of a campaign to build an airplane not just for the Air Force but for all the services. The airplane would have to avoid the death spiral through radical cost-cutting, and it would have to meet the distinctive needs of the Air Force, the Navy, and the Marine Corps with one new machine.

Muellner said he would not like this opportunity one bit. But he knew the perfect candidate—some other general. A few weeks later, Muellner told me early this year, he was called back for a meeting with civilian Pentagon officials, who asked him again how he would like the assignment. He told them, "I have to be honest—I really don't want to lead this program. I have some real misgivings about its likelihood of success." Half an hour after that interview McPeak phoned him. "It didn't work," he said. "You got the job."

As the director of JAST, Muellner was supposed to convert a mere research team into an actual airplane-buying effort—the Joint Strike Fighter project. "From the very beginning," he told me, "the program was resented by all the military services. The new administration had come in and canceled a number of the major programs"—on top of Cheney's A-12 cancellation—"and they really did not have a good solution as to what would replace them."

Through 1994 Muellner assembled a staff and began to consider how to overcome the history of joint-program failure and the current reality of adversaries in every corner of the Pentagon. He met with each of the service chiefs and asked what kind of airplane the services really needed. Admiral Frank Kelso, the Chief of Naval Operations, told him what the Navy required: an airplane with two engines, in case one failed over the ocean; two seats, so that a navigator or a weapons operator could be aboard; and enormous gas tanks, so that the plane could fly 1,000 miles inland from a carrier to launch an attack.

The Air Force? Merrill McPeak told Muellner that the ideal airplane would be a modernized, stealthy version of the F-16, with one seat and one engine, to keep size and weight down and operating costs low. It had to be more durable than the F-16. And one more thing, McPeak said: "I don't want the damned Navy skewing the program to where I can't afford it." The Air Force operated more fighters than the Navy did, so it had more to lose if per-plane costs got out of control.

Speaking for the Marines, Carl Mundy, the commandant, said he wanted the same airplane McPeak wanted—but it also had to land vertically, like a helicopter.

"It became immediately obvious that we were going to have a difficult time with any design that would satisfy all of them," Muellner told me. The Royal Navy's interests also had to be considered. Its aircraft carriers had smaller decks and airplane elevators than the U.S. Navy's, and the plane would have to fit them, too.

The solution Muellner and his team devised was the first major break for the JSF. They tried to cope with inter-service rivalry through a rotating command structure. The directorship of the JSF program, a military position, would switch every two years among the services: an Air Force general would be replaced by a Marine general or a Navy admiral, who would be replaced by another Air Force general. (The rotation reflected the project's nature as a joint Air Force-naval services undertaking.) Each military program director would report to a civilian supervisor from the other service: when an Air Force general was running the program, he would answer to the assistant secretary of the Navy, and vice versa.

Inter-service cooperation could also lead to lower costs. The things that make an airplane expensive are its engine, its avionics (radar and other electronic systems), and the software to run them. These are usually invented from scratch for each new model. What if the components of the different planes were kept as similar as possible, no matter how different the planes might look from the outside?

This concept eventually led to a goal of "80 percent commonality" among the family of airplanes that would meet the three services' needs. The commonality was also designed to keep support costs down in the long run. If a damaged Navy airplane had to land at an Air Force base, all the spare parts would be there. The next time British and American forces fought together, each could support the other's planes.

All this new thinking was in the service of an even blander-seeming, yet revolutionary, concept within the military: CAIV, or "cost as independent variable." In layman's terms this is nothing more than the idea that the airplane would be held to a cost target. In most military contracting, cost is the dependent variable. The Air Force decides, midway through a development project, that an airplane must carry 2,500 pounds of bombs rather than 2,000, and the cost ratchets up accordingly. For the JSF, costs were supposed to be fixed. If one of the services wanted a new feature, it had to take something else out.

When George Muellner gave his first speech after taking command of the joint advanced strike technology program, he drew a line through the "advanced" and changed it to "affordable." "Affordability is absolutely the centerpiece of this program," Leslie Kenne, an Air Force general who was the JSF program director in the late 1990s, told Aviation Week & Space Technology. (Kenne was the first woman to become a three-star general in the Air Force.) "I think this is the first [program] to have cost in the warfighters' requirements document." Translation: "the warfighters," also known as customers, are the military services; a "requirements document" lays out the formal specifications the airplane must meet. Requirements had usually been confined to speed, payload, and so on. Now they would include cost. What a thought! "All of a sudden," Kenne said, the services "are just as accountable for the cost as they are for all the performance, survivability, and lethality that they want. And that makes you behave differently." One of the first cost-driven decisions was that the Navy could get along with a single-engine airplane. Modern engines were reliable enough, and with two engines the cost target would be a joke.

A Jet Like a Helicopter

By 1995 the competition was on. The JSF program office gave grants of about $30 million each to four companies, to produce designs of the new airplanes. (Northrop Grumman, the fourth company, joined the McDonnell Douglas team shortly thereafter.) The companies had to use just the grant money, and not spend any of their own. This was meant in part to keep the largest company, Boeing, from outspending the others. It was also meant to spare all the companies the woes that had attended the B-2 and F-22 programs when bidding contractors invested money in the expectation of huge production runs that never occurred.

The approaches taken by the remaining companies were distinct from the start, primarily because of how each company planned to deal with the Marine Corps. McDonnell Douglas seemed to be betting that sooner or later the military would abandon the fantasy of making an affordable first-rate fighter plane that could also behave like a helicopter. The McDonnell Douglas plane was sleek and attractive, and very effective in its Air Force and Navy roles. But its solution to the Marines' demand was an extra jet engine, with its nozzle pointing downward, which the pilot would switch on when he came in to land. Nothing like this had been tried in any other flying machine, and the approach was sure to be considered high-risk.

The gamble that the Marines would be overruled by the other services was not irrational. The STOVL model would always be technically harder to develop and more risk-prone than the others, and thus more expensive. Because the Marine Corps plane represented only 600 of the planned run of 3,000 JSFs for the U.S. and UK markets, the military might drop it rather than let it delay or distort the cost of the overall program. (The Air Force plans to buy at least 1,750 JSFs, the Navy 480, the British military 150.)

What this calculation left out was the intensity of the Marine Corps' desire for a new STOVL plane of its own, and the deftness of its lobbying skills with Congress and the media. Here the Corps was unique among the services. The Air Force leadership lives and dies to defend the high-end F-22. And "when the Navy thinks of 'platforms,' it thinks of ships," Rudy de Leon, who was the deputy secretary of defense at the end of the Clinton Administration and is now an executive with Boeing, told me. "The airplane is [just one] part of the carrier." If the carrier uses F-18s rather than JSFs, so be it. "But for the Marines, this is their future." Another civilian observer told me, "The other services were given something that fit their needs. The Marines were given something that fit their vision." For reasons that range from a reputation for plain-spokenness to the network of Marines working as military aides in congressional staff offices to the history and aura of the Corps, the smallest of the services has by far the best track record in defending budgets for its projects. By the time of the first downselect, in late 1996, building a STOVL model still mattered, and McDonnell Douglas's odd, risky airplane was squeezed out.

The remaining two planes were odd too. Boeing decided to take a proven route, by modifying the design of the Harrier. The Harrier works on the "direct lift" principle: when it wants to go up like a helicopter, some of the exhaust from its jet engines is diverted from the tail to little downward-pointing nozzles under the wings and the nose. Boeing used an updated system of direct lift. In the crucial transition phase of flight, as the Marine pilot swooped toward, say, a remote base and began descending to a tiny clearing, a trapdoor would close in the engine's tailpipe and the exhaust would be shunted to nozzles in other parts of the plane, cushioning its descent.

The decision to go with direct lift had an engineering consequence and an aesthetic one, and it is hard to tell which was more important. The engineering problem was that the Boeing plane, like the Harrier before it, was chronically on the verge of not being able to lift itself. The airplane weighed roughly 30,000 pounds, so to ascend vertically it needed more than 30,000 pounds of jet thrust. The engine produced that—but barely, and with what were called environmental side effects. For instance, grass would catch fire, or chunks of asphalt would be blasted away. It was like launching a small rocket, and the margin for error was tiny. If the engine were any weaker, it wouldn't lift the airplane; any stronger, and it would burn up the landing zone or carrier deck.

To keep the airplane balanced in vertical takeoffs and landings, its heaviest element, the engine, had to be at the center of the fuselage. (If it were in the tail, the normal place, it would pull the plane down tail first.) This made the body of the airplane look barrel-shaped and chubby—the aesthetic consequence. But it did something far uglier. To get enough air into the engine, and for reasons relating to the stealthy design, the Boeing plane had an intake at the front that looked like an oversized, gaping mouth. In the mid-1990s the plane was called a flying frog or a pelican. By the late 1990s it was known inside the Pentagon as "Monica." When Boeing produced promotional videos of the plane, for an advertising blitz before the ultimate downselect, it hired digital artists who had worked on Titanic to substitute another plane's outline for Monica's. The switch was defensible, because Boeing had been clear about its intention to alter parts of the design—though not the mouth, which was still visible in the video—if it won.

No one with a prominent and official role in the JSF program will say that a factor as frivolous as looks made the slightest difference in the outcome. Darleen Druyun, a powerful civilian official at the Pentagon, ridicules the idea that "emotion" would color a multi-billion-dollar decision. "I've had a number of officers come down and say, 'Druyun, have you looked at the airplane?'" she told me, referring to a previous case in which she favored an "ugly" model. "I said, 'Yes, I've looked at it, and I've flown in it, and I want to tell you something. Looks can be deceiving.'" Mike Hough, a boisterous Marine Corps general who was directing the JSF program when the final decision was made, sputtered when I asked if Boeing's plane was too ugly to win. "Looks made the difference? Not even goddamn close! I want to tell you about the old bottom feeder"—a catfish image for the Boeing plane. "It performed flawlessly. It was incredible, the performance of that airplane. It was an amazing machine, and it still is." (The other, though, was even better.)

People without official roles in the process are not so sure. Manfred Von Nordheim, a dashing German who directs the U.S. office of the European aerospace consortium that builds the Airbus, the Eurofighter, and other products, says: "To me it was obvious. If you looked at the two planes, that one was not going to fly. It was like the Edsel. You wonder, didn't someone say, Wait a minute! This doesn't look right?" Helpful as my contacts at Boeing were, no one was eager to claim credit for the design of the plane.

On the outside Lockheed Martin's entry looked like a normal, jazzy fighter plane. Its oddest feature was inside. In the early 1990s Paul Shumpert and Paul Bevilaqua, two engineers at the company's research site in the remote Mojave Desert of California, did the research leading to a patent for a wholly novel approach to the vertical-lift problem. Their solution had never been put into practice. But if it worked, it would give the Lockheed Martin airplane an enormous edge. In what everyone involved recognized as a "bet the company" decision, Dain Hancock, the president of the aeronautics division, approved an all-out effort to make the new "lift fan" propulsion system work. Hancock and his colleagues believed that if it succeeded, they had a very good chance of winning the JSF competition. If it failed, they would be out of the running. And because it might be decades before the next fighter program was launched, they could be out of the fighter business for good.

The lift fan was essentially a means of harnessing some of the power of the jet engine and using it to lift the plane straight up. The Air Force and Navy versions of the Lockheed Martin plane would have normal jet turbine engines. But in the Marine Corps version a shaft would be connected to the front of the engine and would spin, like the engine's turbines, at thousands of revolutions per minute. The patent for the lift fan covered a system for connecting this whirling shaft, through clutches and a gearbox that joined two shafts at a ninety-degree angle, to a very large fan that would take in air from vents at the top of the plane and then whoosh it out the bottom. This air would provide a cushion on which the plane hovered.

The lift system sounds unlikely as described, and it looked quite Rube Goldbergian in diagrams. It also included a device that seemed to come straight from a Tim Burton movie: a tailpipe that could be made to snake around from its normal straight-back position to point down and support the rear of the plane, driving vertical ascents and descents. Through a kind of alchemist's magic, the same sort of engine that was laboring to hold up the Boeing model could produce a comfortable margin of lifting power for Lockheed Martin. The explanation was that a large volume of air, moved at relatively slow speeds by the fan, could provide a large amount of lifting force. The practical significance of the approach was that it would give the airplane much more flexibility and allow for greater thrust—all of which meant the plane could operate with heavier loads, in worse circumstances, with greater confidence of actually being able to fly straight up. Also, since the air shooting out beneath the plane was at normal atmospheric temperature and was not a jet blast, it would not be setting runways on fire or melting carrier decks.

All that would be true if the never-before-tried fan worked. Through the first few months of tests gearboxes exploded and drive shafts snapped and oil spewed everywhere. But Lockheed Martin had made its choice, and it kept hoping for the best.

Shaping Up the Skunk Works

As the JSF competition intensified in the late 1990s, the only thing that mattered as much as the success of the lift fan was the opinion of Darleen Druyun. Druyun's official title is principal deputy assistant secretary of the Air Force for acquisitions and management. Through the past decade she has been one of the most important civilians at the Pentagon, and she is clearly the most influential woman. She started working as an Air Force contract negotiator more than thirty years ago, soon after she graduated from Chaminade University, in Hawaii, and she has overseen government projects ever since. A former colleague described Druyun to me as "embodying the bureaucratic approach, in the positive sense of the term"—that is, believing in procedure, facts, rules. Her detractors describe her as haughty and cold. When NASA had budget and management emergencies after the Challenger explosion, Druyun was dispatched to be its chief of staff. She returned to the Pentagon in her current job at the beginning of the Clinton Administration, and no signs suggest that she is about to leave. "There is a category of savvy managers in the government that industry lives in fear of," a veteran Pentagon official told me. "That is her. People don't take her on because she is too tough to take on." Another man says, "People live in fear of her, because she not only speaks her mind but they know that in any argument on the merits, she will prevail over the political people. She always knows more."

Darleen Druyun is in her mid-fifties, tall and long-limbed, with dark-brown hair and eyes. When I met her in her office, my strong impression was of an experienced principal at a first-rate public high school who had heard it all and was not taking any guff. She talks in a deliberately patient way, just short of condescending, punctuating her statements with "Okay?" as if challenging the listener to disagree.

In the two years leading up to the final downselect Druyun was the civilian in charge of the JSF. Obviously, many others played important roles. Jacques Gansler, the No. 3 civilian official at the Pentagon in the late 1990s, supervised all weapons programs and paid close attention to this one—the largest. William Cohen, the Secretary of Defense during Clinton's second term, intervened to defend the JSF's potential when congressmen or others were skeptical. But day by day Druyun was on the phone to Boeing and Lockheed Martin, marking their progress against checklists and taking them to task if they were heading over budget.

"We set up these special computer networks," she told me. "Literally from my computer on my desk I can dial right in and look down at every single piece of paper they have. Every single management report. Nothing is hidden from us, okay? It's called full disclosure. There's nothing magical about program management. A lot is just common sense. You don't want surprises. If there is going to be bad news out there, okay, let's float it out. Let's be rational and figure out what is the issue, and how to solve the issue."

Druyun had ample opportunity to apply this philosophy in 1998 and 1999. By then each of the companies had received more than $1 billion to build two flying demonstration models of its airplane. Boeing said that it would have the models done on time—but if it won the contract, the airplanes it actually built would look different, because it had discovered problems with the shape of the wing and the tail. Within the terms of the competition this was acceptable, because the first two models were required only to demonstrate that they could fly and hover in certain ways. Boeing claimed that in the ways that really mattered, its models would be more representative than Lockheed Martin's, because Boeing was building them with the same new and advanced production equipment it would use for the actual fleet, whereas Lockheed Martin was building its two models on its regular equipment. (Lockheed Martin built working models of the production equipment it would use if it won.) Nonetheless, the planned design change didn't help Boeing's prospects.

Lockheed Martin's problems, apart from the tricky lift fan, fell under the general heading of bad management, as the company's officials now concede. Through 1999 Lockheed Martin's management disarray worsened, and Darleen Druyun seemed increasingly exasperated with the company's officials, including its CEO, Vance Coffman. At one bitter meeting with Lockheed Martin officials in 1999 Druyun, according to a widely circulated memo by someone present, said that the company had lost an earlier contract for a satellite because its design was "crappy." She warned them against any funny business with the JSF cost figures. "If I detect BS," she warned, "you go to the bottom of the chart."

The most acute problem was the company's apparent inability to control costs. The two flying models were being developed at the Mojave Desert research site, the "skunk works," which had operated for decades as an independent fiefdom. Under its creator and former director, the renowned designer Kelly Johnson, the assumption had been that the corporate masters left the skunk works alone, and every so often it came up with brilliant new airplane ideas. Which it did: the U-2 and SR-71 spy planes both came from Lockheed's skunk works. No one there was accustomed to answering precise accounting questions from corporate headquarters—let alone having a Pentagon official tap into the books from her computer.

"There was one point [in 1999] where it was strongly urged that we terminate Lockheed and give the program to Boeing," Jacques Gansler told me recently. "Even if to do that we had to give Boeing the Lockheed Martin design to build. The skunk works had just lost cost control. The management was in terrible shape."

Druyun told me that she had never favored actually pulling Lockheed Martin from the project. "There was some tension about Lockheed's management of this program," she said. "But I also have to tell you there was tension with Boeing as well. The answer was the process that we used. We went to them on a very regular basis. We laid out what the issues were. We were looking at cost overruns with both contractors in the neighborhood of $150 to $200 million each. We sat down with them and said, We have no more money to bring to the table, 'kay? It's yours to fix."

Druyun's partner in delivering the message to the companies was Mike Hough, the military director of the program office in the two years leading to the downselect. Hough and Druyun are the same age, but in most other ways they are opposites. Hough graduated from the Naval Academy in 1969 (in the same class as Tom Burbage) and then became a fighter pilot; he is short, burly, wisecracking, profane. Every few minutes during our interview Druyun would tell me, "You just have to look at the facts ..." At least as often Hough summed up situations by saying, "It was a leadership problem ..."

But Hough's version of the shape-up message to the contractors reinforced Druyun's. "What really happened here was the rules changed," he told me recently. "They had to understand that no longer were we going to say [he shifted to a high, mincing voice] 'Oh, my! You're running out of money? We'll go print some more.' No way! I said, 'Your costs are going up? Well, you better find a way to bring 'em down. Talk with your suppliers, make it work.' People and organizations only change after a 'significant emotional event,' and they went through one, which helped them see they had to change."

Hough and Druyun were not the only ones noticing problems. The Congressional Budget Office warned in 1999 that the program could go 50 percent over its target costs. In another report the General Accounting Office recommended that the program be put on hold for months or years, while complicated technology, including the lift fan, was worked out. A few congressmen proposed hearings on delaying the program.

From William Cohen on down, the Pentagon leadership resisted the idea of a moratorium or an extension. This was not just the normal pork-barrel instinct to protect a program—an instinct that showed up strongly in defense of the ever costlier F-22 and the Marine Corps' crash-prone V-22 Osprey. Gansler, Druyun, and Cohen himself argued that delay would be the first step down the death spiral for the Joint Strike Fighter. And it would leave the Marines and the Air Force with no replacements for their aging fleets.

The Pentagon leadership could see something else as well: after the warnings from Druyun and Hough, the contractors had gotten the message. Lockheed Martin shook up the management of its JSF team, putting Tom Burbage in charge and persuading the skunk works to act like part of a larger organization. The gearbox for the lift fan stopped blowing up. Soon the company had every indication short of a real flight test that the novel lift fan would work. Meanwhile, Boeing concentrated on the advanced manufacturing system that it thought would be the strength of its proposal. Its advertising claimed that the Boeing JSF would start on "Day One, not Square One"—deriding Lockheed Martin's need to create a new production process.

There was one major cost setback. Late in 2001, after months of telling the services that if they wanted any new feature on the plane (more weapons capacity, longer range) they had to give up something else of equal cost, Hough, Druyun, and their colleagues accepted changes in the JSF's design, and in its 10-million-line software coding, that would add about 10 percent to the cost of each airplane. (Why so much software? Because the engine, the weapons system, the elaborate radar, and countless other features are run by computer.) A series of such alterations would start the death spiral; so far, this change has been an isolated case. And both contractors were developing a more accurate understanding of just what the costs were. Late in the process Druyun laid out three sets of projections for the entire cost of the program over the next twenty-five years. One set of figures came from the Pentagon's auditors, and one from each of the two companies. The projected totals differed by less than a tenth of one percent.

Through 2000 the companies put their efforts into actual testing, and they went from strength to strength. Boeing's first test flight was in September, Lockheed Martin's a month later. Each was a success. The following June both companies were ready for the most demanding flight test: the vertical takeoff and landing. By chance each company's most critical test flight took place on the same day, June 24, 2001, on opposite ends of the country. At Patuxent River Naval Air Station, in Maryland, the Boeing airplane made a normal takeoff, and once flying it slowed until it was no longer moving forward and held a hovering position. Near Palmdale, California, Lockheed Martin planned a strictly vertical takeoff to demonstrate the power of its lift fan. The pilot planned to take the airplane only a few inches off the ground. "Everybody was standing there going 'God, I hope it works this time,'" Tom Burbage later told George Wilson, of National Journal. As the pilot pushed in the controls, the plane rose smoothly—all the way to twenty-five feet, where it hovered for thirty-five seconds before coming down. Burbage's team knew that the sensitivity of the controls needed adjustment, but also that the fan had power to spare. In July, Lockheed Martin sent its plane on a tour de force called Mission X. It made a short takeoff, accelerated and broke the sound barrier, and then returned and landed vertically—a combination no airplane had ever performed before.

Through the summer pilots carried out a successful series of flights to provide "test points" showing that the planes had met the program's requirements. "Did they fly it twice a week, like most experimental airplanes?" Mike Hough asked me rhetorically. "Noooo, they flew it three times a day. They flew it five times a day! I had to take the keys away from them. People were stunned. Ordinarily it would take four or five months to get the test points. We had them in twenty days each, Boeing and Lockheed alike."

Hough probably has the evangelist's air by nature, but he had a specific reason for it in early 2001. Objectively, all aspects of the program had been improving through the previous year. Management was straightening out. The hardware and software were functioning as planned. But political support for the JSF appeared to have ebbed. A normal defense program, such as for the building of a ship or a tank, has automatic backing from the military service that "owns" it, and also from politicians in the districts where it will be built. Chuck Spinney, the civilian analyst, has called this military version of pork-barrel spending "political engineering," and notes that most programs deliberately choose suppliers in as many congressional districts as possible, in order to maximize support. The F-18, for example, has subcontractors in every state but Wyoming. But as a joint program, the JSF had all-out support only from the Marine Corps, which couldn't pay for it single-handedly. The plane was in constant peril of being abandoned by an indifferent Navy or stabbed in the back by the Air Force in an effort to save its cherished F-22. The program was expected to have direct suppliers in only half the states even when it was up and running, in part because of extensive partnerships with European companies.

Tom Burbage's team at Lockheed Martin made daily entries on two charts showing their prospects. One was "P-Win" —the probability that Lockheed Martin would win the contract. The company still considers this chart confidential, but clearly the P-Win rating shot up when it became clear that the lift fan actually worked. The other chart was "P-Go" —the probability that the program itself would survive.

During the presidential campaign P-Go was low. If the Democrats won, the Lockheed Martin team reasoned, sooner or later they would cut the defense budget. And if the Republicans won, the JSF could still be vulnerable, because it was a "Democrat" airplane. During the campaign George W. Bush said that the way to improve the military was to "skip a generation" to futuristic weapons, such as remote-controlled unmanned planes. In theory this could be a reason to skip the F-22, or the Navy's latest-model F-18. In reality the argument was more likely to be turned against the JSF.

When Donald Rumsfeld's team moved into the Pentagon, it was not clear whether Rumsfeld's program of "transforming" the military would embrace the JSF, with its lower cost and advanced production systems, or exclude it. At a systems-review briefing in July of last year, Hough seemed to surprise Rumsfeld and his deputy secretary, Paul Wolfowitz, by giving his standard motivational pitch for the JSF rather than calmly analyzing their questions about why this program was necessary and practical. But two months later the country was at war, Rumsfeld and Wolfowitz were preoccupied, and all programs went forward. The P-Go line shot up.

Will It Take Off?

The companies knew that the final selection process, organized by Darleen Druyun, would be thorough to a fault. The 200-plus civilian and military members of the selection panels took oaths of impartiality and lifelong confidentiality. The contestants were judged on three main criteria: affordability, both in the initial cost of the plane and in long-term operating expenses; the performance of the flying models; and each company's record in past programs. Some 500 aspects of each company's program were rated according to these criteria. Darleen Druyun showed me a few of the ranking sheets, and said that the complete results totaled tens of thousands of pages.

While the selection teams pored over the results, the companies couldn't help themselves: they intensified their lobbying and advertising campaigns. Lockheed Martin took one floor of an office building in Crystal City, just across the Potomac from the Capitol, and filled it with hardware and mock-ups of the JSF. Politicians, staffers, foreign dignitaries, and reporters could make appointments to get a briefing and "fly" the plane in a full-scale cockpit replica. Both companies spent heavily on broadcast and print advertisements in the Washington area in 2001. A minor purpose of the ads was to deal with the issue of the plane's appearance. "The New Look of Lethality," an ad would say, or "The New Look of Affordability." "From some angles, [the plane] looks menacing," Thomas Young, the director of the JSF ad campaign for Boeing, told me in St. Louis, after Boeing lost. "We thought it could be an advantage to have an airplane that looks mean." But the major purpose of the ads was to stress "facts," "reality," "delivery." Boeing had already developed the machines that would produce the plane, and Lockheed Martin was only promising that it could. For its part, Lockheed showcased its sleek-looking airplane, with the reminder that it had a brand-new propulsion system that worked.

Why did either company bother, given that Druyun's team knew far more about the airplanes than it could learn from any ad? The rationale was that it could not hurt to remind congressmen, journalists, and the rest of the capital of the virtues of the product. But as in any military program, advertising had other benefits. "It would be safe to say that there was a great desire from many people at the Pentagon for us to advertise and to be out there, as part of selling the program," Doug Kennett, the head of communications for Boeing's military-aircraft division, told me. "A lot of our early advertising was selling JSF, no matter who won."

After the announcement Darleen Druyun told Jerry Daniels, the president of Boeing's military-aircraft division, that his project was rated A. But Lockheed's was A+. Hough likes to say that he originally feared having one candidate that rated four on a ten-point scale and another that rated six. "I got a nine and a ten," he says. Lockheed won for the reason that Dain Hancock and his colleagues had foreseen: the lift fan made the airplane perform better. Boeing's edge in manufacturing could not overcome this edge in design.

What, exactly, has Lockheed Martin won? "They have bought themselves the opportunity to compete, continuously, through twelve straight annual funding cycles to keep this thing alive," one civilian analyst told me. "It's going to be tough, since the airplane is not exactly what anyone wanted, and more-pressing things will come along." One of the new things sure to come along is unmanned combat aircraft. Many members of Boeing's JSF team have been switched to its unmanned-vehicles project. When meeting with several of them, I mentioned a nightmare scenario for Lockheed Martin: that Boeing, while playing the good loser, would get its revenge by successfully promoting unmanned vehicles as the real way to make defense affordable. The Boeing men all laughed when I said this. Of course that is what they have in mind.

The more important question is What, so far, has the country won? The most optimistic interpretation would be that the JSF represents the introduction of the best, real parts of the New Economy to the messy business of building military machines. Talking to Tom Burbage in Washington and in Fort Worth, I kept being surprised by how much he sounded like the high-tech executives I have interviewed in recent years. "You have to create an environment where people can succeed," he would say. "Too often you put people in a situation where they can't really succeed, and when they don't, you hammer them." He talked about inviting bids from foreign suppliers, on the "best athlete" principle. But of course Burbage is a high-tech executive. After Lockheed Martin won the contract, Burbage took seventy team members on a weekend retreat where he showed them a large mock-up cover of Fast Company magazine dated five years into the future. "JSF: Simply the Best!" the cover line read. He asked his associates to think through the steps that would lead to such success. (He considered preparing another cover, "JSF: The Smoking Hole," but didn't go through with it. Fast Company learned about the mock-up, got in touch with Burbage, and ended up doing a real cover story about him this spring.)

In ways that go far beyond retreats and motivational talk, the production of this airplane could be a large step forward. Here are two illustrations, from among dozens of possibilities.

Most military airplanes are handmade to a degree that is hard to believe if you haven't seen it. Putting together a wing for a brand-new, top-of-the-line F-18 involves thousands of rivets and hundreds of parts. The final assembly of the wing takes place on enormous tooling structures that force the various spars, ribs, and skin pieces into proper alignment, so that they can be fastened. This is the way cars were built thirty years ago, and why they rattled. At the Lockheed Martin plant in Fort Worth, under the guidance of Martin McLoughlin, the JSF's director of manufacturing, I saw a demonstration of modern manufacturing systems like the ones already in place at Boeing. The wing is formed as two great halves, and the joined halves are matched so precisely to the fuselage that they snap in. The computer industry as we know it would not exist were it not for high-speed, high-precision assembly, nor could America's car makers compete with Japan's had they not used these techniques. This is the first time these methods will be used for the military.

The second example is stealth. The most expensive military plane ever built is the original stealth bomber, the B-2. Each B-2 cost $1 billion. Today, adding "low observable," or stealthy, qualities to an airplane is practically a no-cost option. So much has been learned about how to design wings, engine intakes, and other features that there is little point in building non-stealthy airplanes.

The JSF also represents the New Economy in that it is a global airplane. Richard Aboulafia, the aviation analyst, says that if the JSF meets cost targets, it might well be the only viable fighter plane on the world market a generation from now. "As a libertarian, I eschew any targeted industrial policy," he told me. "But if I were the U.S. government, I would think this was irresistible. If you make the assumption that this is the same economic class as the F-16, there's no reason you couldn't sell to twenty-five countries, maybe more."

These projections fit the "Simply the Best!" story. But naturally there's another possibility. Even if the program somehow keeps meeting its cost targets, the JSF planes coming into service won't arrive fast enough to replace the planes that are getting old. In the meantime, the United States will spend several hundred billion dollars maintaining a fleet that becomes smaller by the year. And if the program can't meet its targets, everything changes. The export market dries up. The death spiral begins.

The fatalistic view of the JSF's prospects rests not on what has happened to this project so far but on what has happened to virtually every other project the Pentagon has undertaken. Early this year I spoke with a man who has been involved in Defense Department budgeting since the 1960s. In theory the JSF was a good idea, he said. So far it had stayed surprisingly close to its targets. But the weight of history suggests that obstacles are still ahead. Of the dozens of weapons programs this man had observed through their development, how many had gone into service at or under the cost specified when Congress initially approved them? None. How many had met the higher cost target authorized when they moved from early development to actual production? None. How many had gone into service in the numbers foreseen when the plans were approved? None. So if the JSF meets its targets, it will do something no previous weapons program has done. That is a sign of this program's ambition—and of the odds against it.